JP2013034181A - Image processing apparatus and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent degradation of image quality due to isolation of a thinned pixel.SOLUTION: An image processing apparatus 1 comprises a blending processing unit 5 which determines a pixel value G of a target pixel of when a smoothing processing and a thinning processing are realized at the same time depending on a combination of whether the thinning processing is performed or not and whether the smoothing processing is performed or not on the target pixel, and a thinning adjustment unit 6 which determines whether a thinned pixel becomes isolated by change amounts dST in pixel value due to the thinning processing of the target pixel and the surrounding pixel, the thinning adjustment unit resetting the change amount dST in pixel value due to the thinning processing of the thinned pixel among the target pixel and the surrounding pixel adjacent to the target pixel in the edge direction or in an opposite-edge direction to a value of zero if the thinned pixel is determined to be isolated. The blending processing unit 5 uses the reset change amount dST in pixel value due to the thinning processing to determine the pixel value G of the target pixel.

Description

  The present invention relates to an image processing apparatus and an image processing method.

Thinning processing is known as one of image processing. In the thinning process, the sharpness of a character or the like is improved by reducing the pixel value for the edge of the character or the like.
Image processing that targets edges also includes smoothing processing. In the smoothing process, the edge value of the shaded portion can be visually smoothed by increasing / decreasing the pixel value of the edge of the shaded portion of the character or the like to obtain a halftone.

  Although the thinning process and the smoothing process target the same edge, it is difficult to perform the two processes simultaneously because the purposes of the processes are different. For example, if thinning is applied to the straight line part of the character and smoothing is applied to the hatched part, a sharp and smooth character is obtained. However, the structure of the straight line part and the oblique line part is detected, and the thinning process and smoothing are performed according to the detected structure. Such a character is difficult to realize by a method of switching processing (see, for example, Patent Document 1). This is because there are pixels belonging to both the straight line portion and the shaded portion, and an unexpected result occurs, for example, a pixel value is increased by smoothing although thinning is appropriate instead of smoothing.

  On the other hand, the applicant has developed an image processing apparatus that simultaneously realizes a thinning process and a smoothing process (see, for example, Patent Document 2). According to this image processing apparatus, the result of the smoothing process and the thinning process is held as a change amount from the original pixel value. If the pixel value when the retained change amount is added to the original pixel value exceeds the maximum value or becomes a negative value, the excess or insufficiency is the predetermined positional relationship with the target pixel. It is adjusted according to the pixel value of the adjacent pixel. By adjusting the processing results of the thinning process and the smoothing process for not only the pixel of interest but also two adjacent pixels, the two processing results can be easily digested.

JP-A-7-334672 Japanese Patent No. 4225337

  Originally, when a thin line with a small line width is thinned, since both sides of the thin line are thinned, the thin line may fade or disappear. For example, in the case of a thin line with a width of 2 pixels, if the pixel value is reduced from both sides by thinning, the thin line itself disappears. In order to prevent the disappearance of such a thin line structure, conventionally, the thin line structure is determined during the thinning process and excluded from the thinning target.

  However, the exclusion of such a thin line structure may cause unintended results. For example, when image data is irreversibly compressed, the original image data is not completely restored by decompression, and characters and the like may partially deteriorate in image quality. For example, the outline may not be restored smoothly and may be sharply pointed. A sharp portion is determined to be a thin line structure during thinning processing, and thinning is invalidated only for that portion. Since there are a portion that is not thinned and a portion that is thinned in the same character, the line width of the character is unbalanced.

  In addition, the oblique lines close to the horizontal have a structure in which long horizontal lines are arranged in a staircase pattern. However, since the line width temporarily increases in the stepped portion, only the stepped portion is thinned although the horizontal line is not subject to thinning. The phenomenon that occurs. While there are many horizontal line pixels that are not thinned, the thinned stepped portion is 1 or 2 pixels at most, so the thinned portion is isolated. If the thinning process and the smoothing process are simultaneously performed on an image having such an isolated thinned portion, the pixel value is excessively reduced only in the thinned portion, and neither thinning nor smoothing can be realized, and the image quality is improved. to degrade.

  An object of the present invention is to prevent image quality deterioration due to isolation of thinned pixels.

According to the invention of claim 1,
A thinning processing unit for thinning the image data;
A smoothing processing unit for performing a smoothing process on the image data;
An edge direction determination unit that determines an edge direction for each pixel of the image data;
Using the pixel value of the target pixel, each pixel value of the peripheral pixel of the target pixel, the amount of change of the pixel value by the smoothing process, the amount of change of the pixel value by the thinning process, and the edge direction determined for the target pixel, The blending processing unit that determines the pixel value of the target pixel when the smoothing processing and the thinning processing are realized at the same time according to the combination of the presence or absence of thinning and the presence or absence of smoothing;
It is determined whether or not the thinned pixel is isolated based on the amount of change in the pixel value of the target pixel and its surrounding pixels by the thinning process. If it is determined that the pixel is isolated, the target pixel or the edge direction of the target pixel Or a thinning adjustment unit that resets a change amount of a pixel value due to the thinning process of a thinned pixel among neighboring pixels adjacent in the reverse edge direction to a value of 0,
The blend processing unit is provided with an image processing apparatus that uses a pixel value change amount by the thinning process that has been reset to determine a pixel value of the target pixel.

According to invention of Claim 2,
When the thinned pixel is a peripheral pixel adjacent in the edge direction or the reverse edge direction of the target pixel, the thinning adjustment unit thins the peripheral pixel on the condition that the target pixel is thinned. The image processing apparatus according to claim 1, wherein the change amount of the pixel value due to is reset to a value of 0.

According to invention of Claim 3,
The thinning adjustment unit is configured so that the blending processing unit determines whether the pixel of interest has thinning or smoothing according to the amount of change in the pixel value of the pixel of interest by the thinning processing and the amount of change of the pixel value by the smoothing processing. The image processing apparatus according to claim 1, wherein the reset is performed after classifying the combinations.

According to invention of Claim 4,
The image processing apparatus as described in any one of Claims 1-3 provided with the control part which switches whether the said reset by the said thinning adjustment part is performed.

According to the invention of claim 5,
A process of thinning the image data;
Applying a smoothing process to the image data;
Determining an edge direction for each pixel of the image data;
Using the pixel value of the target pixel, each pixel value of the peripheral pixel of the target pixel, the amount of change of the pixel value by the smoothing process, the amount of change of the pixel value by the thinning process, and the edge direction determined for the target pixel, The step of determining the pixel value of the pixel of interest when the smoothing process and the thinning process are realized simultaneously according to the combination of the presence or absence of thinning and the presence or absence of smoothing;
It is determined whether or not the thinned pixel is isolated based on the amount of change in the pixel value of the target pixel and its surrounding pixels by the thinning process. If it is determined that the pixel is isolated, the target pixel or the edge direction of the target pixel Or resetting the amount of change in the pixel value due to the thinning process of the thinned pixel among neighboring pixels adjacent in the reverse edge direction to a value of 0,
The step of determining the pixel value of the pixel of interest is provided with an image processing method for determining the pixel value of the pixel of interest using the reset amount of the pixel value caused by the thinning process that has been reset.

  According to the present invention, when the pixel value is determined when the thinning process and the smoothing process are realized simultaneously, it is possible to invalidate the thinning of the isolated thinned pixel. By disabling thinning, it is possible to prevent deterioration in image quality caused by the thinning.

It is a block diagram of the image processing apparatus which concerns on this Embodiment. It is a figure which shows the area | region of 7x7 pixel. It is a block diagram of a smoothing process part. It is an example of the template used for the pattern detection of the edge which should be smoothed. It is a block diagram of a thinning process part. It is an example of determination of an edge direction. It is an example of determination of an edge direction. It is an example of determination of an edge direction. It is a block diagram of a blend process part. It is a flowchart of the process which a classification | category part performs. 5 is a flowchart of processing performed by a pixel value determination unit in case 1; (A), (b) is a figure which shows the positional relationship of the attention pixel assumed in the case of case 1 and a surrounding pixel. 10 is a flowchart of processing performed by a pixel value determination unit in case 2; (A), (b) is a figure which shows the positional relationship of the attention pixel assumed in the case of case 2, and a surrounding pixel. 10 is a flowchart of processing performed by a pixel value determination unit in case 3; (A), (b) is a figure which shows the positional relationship of the attention pixel and surrounding pixel assumed in the case of case 3. FIG. 10 is a flowchart of processing performed by a pixel value determination unit in case 4; (A), (b) is a figure which shows the positional relationship of the attention pixel and surrounding pixel assumed in the case of case 4. FIG. (A) It is an image figure of the process which a blend process part performs. (B) It is a figure which shows the processing result of a blend process part. (A) It is a figure which shows the original image. (B) It is a figure which shows a target image when performing a thinning process and a smoothing process simultaneously. (C) It is a figure which shows the process result which performed only the smoothing process to the original image. (D) It is a figure which shows the process result which performed only the thinning process to the original image. (E) It is a figure which shows the process result at the time of switching a smoothing process and a thinning process according to the structure of an image. (F) It is a figure which shows the process result at the time of adjusting a pixel value small in a smoothing process. It is a flowchart of the process which a thinning adjustment part performs. (A) The original image. (B) An image after smoothing processing of the original image. (C) An image after thinning processing of the original image. (D) An image in which a thinning process and a smoothing process are simultaneously performed on the original image. (E) An image after thinning processing when thinning of isolated thinned pixels is invalidated. (F) An image in which thinning processing and smoothing processing are simultaneously performed on the original image after thinning is disabled.

  Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a configuration of an image processing apparatus 1 according to the present embodiment.
As shown in FIG. 1, the image processing apparatus 1 includes a smoothing processing unit 2, a thinning processing unit 3, an edge direction determination unit 4, and a blend processing unit 5.
The image processing apparatus 1 is provided with a line memory for a total of 9 pixels from the 0th line to the 8th line in the sub-scanning direction, and image data is input and held in this line memory while being shifted by one pixel in the main scanning direction every clock. Is done. The image data is 8 bits, and the range of pixel values is 0-255. In the case of a color image, image data composed of four colors Y (yellow), M (magenta), C (cyan), and K (black) is input.

The smoothing processing unit 2 performs a smoothing process on each pixel of the image data. The smoothing process is also called anti-aliasing.
The smoothing process is performed in units of 7 × 7 pixels. The smoothing processing unit 2 holds 7 lines of image data from the line memory for 7 clocks, that is, 7 pixels in the main scanning direction to obtain 7 × 7 pixels.

Three smoothing processing units 2 are provided. The upper smoothing processing unit 2 has 0 to 6th lines, the middle smoothing processing unit 2 has 1st to 7th lines, and the lower smoothing processing unit 2 has 2 to 8th lines. The image data of the eyes are input respectively.
FIG. 2 shows an area of 7 × 7 pixels input to the smoothing processing unit 2 in the middle stage. In FIG. 2, identification numbers 01 to 49 indicating pixel positions are assigned to the respective pixels.

  Each smoothing processing unit 2 performs a smoothing process on a pixel located at the center of 7 × 7 pixels, and calculates and outputs a pixel value change amount dSMT by the smoothing process. That is, a pixel value of 1 × 7 pixels is input to each smoothing processing unit 2 every clock, and a change amount dSMT of a pixel value of 1 × 3 pixels (identification numbers 18, 25, and 32 pixels) is output.

As shown in FIG. 3, each smoothing processing unit 2 includes an edge determination unit 21, a template determination unit 22, a template selection unit 23, and a pixel value calculation unit 24.
Each smoothing processing unit 2 is different in the position of the pixel to be processed, but the processing content is the same. Therefore, the smoothing processing will be described with the smoothing processing unit 2 in the middle stage that performs the smoothing processing on the pixel of interest having the identification number 25 as a representative.

The edge determination unit 21 determines whether or not there is an edge included in the 7 × 7 pixel. This is because an object to be smoothed by the smoothing process, that is, a pixel whose pixel value is changed is an edge of a character or a line drawing.
The edge determination unit 21 calculates a difference in pixel value between two adjacent pixels, and determines that there is an edge between the two pixels if the difference is equal to or greater than a threshold th1. If the difference is less than the threshold th1, the edge determination unit 21 determines that there is no edge between the two pixels. th1 is a predetermined threshold for edge determination. The difference is calculated as an absolute value, and the two pixels for which the difference is calculated are adjacent pixels in a predetermined positional relationship.

The following is a combination of two pixels A and B for which the difference is calculated, and the combination is indicated by the identification number of the pixel A−the identification number of the pixel B.
03-10, 04-11, 05-12, 10-17, 10-18, 11-17, 11-18, 11-19, 12-18, 12-19, 15-16, 16-17, 16- 24, 17-18, 17-23, 17-24, 18-19, 18-24, 18-25, 18-26, 19-20, 19-26, 19-27, 20-21, 20-26, 22-23, 23-24, 23-31, 24-25, 24-30, 24-31, 24-32, 25-26, 25-32, 26-27, 26-32, 26-33, 26- 34, 27-28, 27-33, 29-30, 30-31, 31-32, 31-38, 31-39, 32-33, 32-38, 32-39, 32-40, 33-34, 33-39, 33-40, 34-35, 38-45, 39-46, 40- 7

  The edge determination unit 21 outputs V [AB] = 1 when it is determined that there is an edge, and outputs V [AB] = 0 when it is determined that there is no edge. In V [AB], A and B indicate two pixels used for calculating the difference, and the values of the identification numbers 01 to 49 of the pixels A and B are input. For example, if it is determined that there is an edge between the pixel A and the pixel B from the difference between the pixel A with the identification number 03 and the pixel B with the identification number 10, V [0310] = 1 is output.

  The number of combinations of pixels A and B is 56 in total, and the arithmetic circuit must obtain a total of 56 subtraction results in one clock. However, the number of subtractions can be reduced by optimizing the arithmetic circuit. The arithmetic circuit outputs the calculation result of the pixel of interest every clock, but the pixel of interest moves to the right by one pixel every clock. Therefore, the pixels A and B subtracted with the current clock become pixels A-1 and B-1 with the next clock. At this time, the subtraction result V [AB] used in the current clock becomes the same as the value of V [(A-1) (B-1)] in the next clock. When one clock elapses, the subtraction result V [(A-1) (B-1)] becomes the same value as V [(A-2) (B-2)]. By using such characteristics, it is possible to divert the subtraction result held as the subtraction result of the same line by holding the subtraction result for 6 clocks.

The following shows that a certain subtraction result can be used as the subtraction result written to the left by the next clock or subsequent clocks. According to this, since one subtraction result is held for 6 clocks, it can be used as a maximum of 6 subtraction results. By adopting such a circuit configuration for diversion, the subtraction required 56 times can be reduced to substantially 17 times.
V [0310] ← V [0411] ← V [0512]
V [1017] ← V [1118] ← V [1219]
V [1018] ← V [1119]
V [1117] ← V [1218]
V [1516] ← V [1617] ← V [1718] ← V [1819] ← V [1920] ← V [2021]
V [1624] ← V [1826] ← V [1927]
V [1723] ← V [1824] ← V [2026]
V [1724] ← V [1825] ← V [1926]
V [2223] ← V [2324] ← V [2425] ← V [2526] ← V [2627] ← V [2728]
V [2331] ← V [2432] ← V [2634]
V [2430] ← V [2632] ← V [2733]
V [2431] ← V [2532] ← V [2633]
V [2930] ← V [3031] ← V [3132] ← V [3233] ← V [3334] ← V [3435]
V [3138] ← V [3239] ← V [3340]
V [3139] ← V [3240]
V [3238] ← V [3339] V [3845] ← V [3946] ← V [4047]

The template determination unit 22 compares V [AB] output from the edge determination unit 21 with the template and determines a matching template. The template is designed so that a pattern of edges to be smoothed can be detected.
FIG. 4 is a template as an example. In FIG. 4, the numbers in parentheses are identification numbers assigned to the templates.

  As shown in FIG. 4, the pixels on the template are classified into three, a, b, and c. The template determining unit 22 corresponds to V [AB] = 0 between pixels corresponding to the classification a on the template, V [AB] = 0 between pixels corresponding to the classification b, and corresponds to the pixels corresponding to the classification a and the classification b. When V [AB] = 1 between pixels to be matched is determined, it is determined that the template matches. The pixel of classification c does not matter whether there is an edge. There may be a case where a plurality of templates are matched instead of only one, and the template determination unit 22 compares all the templates and detects all the matching templates.

  For example, when all of V [1824] = 0, V [2632] = 0, V [1826] = 1, V [1819] = 0, and V [3132] = 0 are satisfied, the template determination unit 22 identifies the identification number. It is determined that the template matches the 21 template. Further, when all of V [2324] = 0, V [3031] = 0, V [1920] = 0, and V [2627] = 0 are satisfied, the template determination unit 22 not only has the identification number 21 but also the identification number. It is determined that the template 213 also matches.

  The template selection unit 23 selects one template from one or a plurality of matched templates by template matching of the template determination unit 22. Priorities are set for the templates shown in FIG. According to this priority order, the template selection unit 23 selects a template with the highest priority order from the matched templates. The priority order of the templates is higher in the order of identification numbers 24, 23, 22, 21, 233, 213, 243, 223, 252, 251, 254, 253, 262, 261, 264, 263.

  The pixel value calculation unit 24 uses the pixel values of the pixels corresponding to the classifications a and b defined in the template selected by the template selection unit 23 to calculate the change amount dSMT of the pixel of interest by the smoothing process. ,Output.

The formula for calculating dSMT is as follows. In the calculation formula, C is the pixel value of each pixel indicated by the identification number in []. The pixel value of the target pixel is indicated by C [25].
When the identification number of the selected template is 21, 22, 23, 24,
dSMT = (C [18] + C [25] + C [32]) / 3-C [25]
When the identification number of the selected template is 213, 233,
dSMT = (C [18] × 2 + C [25] + C [32] × 2) / 5−C [25]
When the identification number of the selected template is 223, 243,
dSMT = (C [24] × 2 + C [25] + C [26] × 2) / 5−C [25]

When the identification number of the selected template is 251, 252,
dSMT = (C [25] × 4 + C [32]) / 5−C [25]
When the identification number of the selected template is 253, 254,
dSMT = (C [25] × 4 + C [18]) / 5−C [25]
When the identification number of the selected template is 261,262,
dSMT = (C [25] × 4 + C [26]) / 5−C [25]
When the identification number of the selected template is 263, 264,
dSMT = (C [25] × 4 + C [24]) / 5−C [25]

  By adding the obtained change amount dSMT to the pixel value of the target pixel, the pixel value of the target pixel after the smoothing process is obtained. According to such smoothing processing, a pixel of interest corresponding to an edge to be smoothed can be detected based on whether the difference in pixel value between two adjacent pixels is substantially the same value or a different value. Since the detected pixel value of the target pixel is converted to an average value obtained by weighting the pixel values of the target pixel and its surrounding pixels, smoothing is possible regardless of the foreground and background pixel values including halftones. Can be processed.

The thinning processing unit 3 performs thinning processing on each pixel of the image data.
The thinning process is performed in units of 3 × 3 pixels. The thinning processing unit 3 inputs 3 lines of image data for 3 clocks from the line memory to obtain 3 × 3 pixels. Three thinning processing units 3 are provided. The upper thinning processing unit 3 includes the second to fourth lines, the middle thinning processing unit 3 includes the third to fifth lines, and the lower thinning processing unit 3 includes the third thinning processing unit 3. The image data of the 4th to 6th lines are input. Each thinning processing unit 3 performs thinning processing on a pixel located at the center of 3 × 3 pixels, and calculates and outputs a pixel value change amount dST by the thinning processing. That is, a pixel value of 1 × 3 pixels is input to each thinning processing unit 3 every clock, and a change amount dST of a pixel value of 1 × 3 pixels (identification numbers 18, 25, and 32 pixels) is output. .

As shown in FIG. 5, each thinning processing unit 3 includes an edge strength calculation unit 31, an edge determination unit 32, a thinning determination unit 33, and a pixel value calculation unit 34.
Each thinning processing unit 3 is different in the position of the pixel to be processed, but the processing content is the same. Therefore, the thinning processing will be described with the thinning processing unit 3 in the middle stage that processes the pixel of interest having the identification number 25 as a representative.

The edge strength calculation unit 31 obtains the edge strength at the target pixel. The edge strength calculation unit 31 calculates differences E0 to E7 between the target pixel and its surrounding pixels, and obtains the maximum value PED among them. The calculation formulas for E0 to E7 are as follows. C is the pixel value of each pixel indicated by the identification number in [].
E0 = C [25] -C [17]
E1 = C [25] −C [18]
E2 = C [25] -C [19]
E3 = C [25] −C [24]
E4 = C [25] -C [26]
E5 = C [25] -C [31]
E6 = C [25] -C [32]
E7 = C [25] -C [33]

The edge strength calculation unit 31 inverts the signs of E0 to E7 and obtains the maximum value RED among them. The thinning processing unit 3 corrects PED = 0 if PED <RED, and RED = 0 if RED <PED.
E0 to E7, PED, and RED obtained as described above are edge strengths.

In the case of a color image, the edge determination unit 32 obtains TPED and TRED by the following formula using PED and RED obtained from the image data of each color. In [], y represents yellow, m represents magenta, c represents cyan, and k represents edge intensity obtained from image data of black color.
TPED =
(PED [y] × Wy + PED [m] × Wm + PED [c] × Wc + PED [k] × Wk)
TRED =
(RED [y] × Wy + RED [m] × Wm + RED [c] × Wc + RED [k] × Wk)
In the above equation, Wy, Wm, Wc, and Wk are predetermined coefficients, and Wy + Wm + Wc + Wk ≦ 255.
In the case of a monochrome image, TPED = PED [k] × Wk and TRED = RED [k] × Wk.

  TPED and TRED are values obtained by weighting and adding edge strengths PED and RED of each color with coefficients Wy, Wm, Wc, and Wk corresponding to specific luminous sensitivity, and the visual density when each color is overlaid. Can be an indicator. Since the thinning target is the foreground edge, the edge determination unit 32 determines which of the target pixel and its surrounding pixels is the foreground and which is the background by determining the magnitude relationship between TPED and TRED. If TPED> TRED, the edge determination unit 32 determines that the pixel of interest is a foreground edge having a higher density than the background, and outputs ED = 1. On the other hand, if TPED ≦ TRED, the edge determination unit 32 determines that the pixel of interest is not a foreground edge, and outputs ED = 0.

  The thinning determination unit 33 determines whether or not the target pixel determined to be the foreground edge can be a target for thinning. Although it is an edge, if the pixel of interest does not have a pixel value sufficient for thinning, that is, to reduce the pixel value, the pixel value becomes very small or disappears because the pixel value is reduced, and the image is interrupted. , It must be excluded from thinning. The thinning determination unit 33 compares the pixel value C [25] of each color of the target pixel with the threshold Th2, and if the pixel value C [25] is equal to or greater than the threshold Th2 in any color, the thinning determination unit 33 is sufficient for thinning. It determines with having a pixel value, and outputs SS = 1. On the other hand, if the pixel value C [25] of the pixel of interest is less than the threshold Th2 in all colors, the thinning determination unit 33 is not subject to thinning because the pixel of interest does not have a pixel value sufficient for thinning. And SS = 0 is output.

The pixel value calculation unit 34 calculates and outputs the change amount dST of the pixel value of the target pixel due to the thinning process. When ED = 1 and SS = 1, since the target pixel is an edge and has a pixel value sufficient for thinning, the pixel value calculation unit 34 thins the target pixel and changes the pixel value change amount dST by the following equation. Is calculated. The change amount dST is obtained for each color using PED and RED obtained for each color.
dST = (PED-RED) × STVL
The STVL is a coefficient for adjusting the thinning intensity and is predetermined for each color. STVL can take a range of 0-1.

  The pixel value after the thinning process is obtained by adding the obtained change amount dST to the original pixel value. As can be seen from the calculation formula of dST, in the thinning process, dST becomes a negative value and the pixel value of the pixel of interest is reduced to literally make the line thin, and dST becomes a positive value. In some cases, the pixel value is increased and thickened.

  When ED = 1 and SS = 1 are not satisfied, the pixel of interest is not an edge, or is an edge but does not have a pixel value sufficient for thinning, so the pixel value calculation unit 34 does not thin the pixel of interest, A pixel value change amount dST = 0 is output.

The edge direction determination unit 4 determines the edge direction in each pixel of the image data.
The determination of the edge direction is performed in units of 5 × 5 pixels. The edge direction determination unit 4 inputs image data of the second to sixth lines from the line memory for five clocks, and obtains 5 × 5 pixels centered on the target pixel in the blend processing unit 5. The edge direction determination unit 4 determines and outputs the edge direction pos for the target pixel located at the center of the input 5 × 5 pixels. That is, a pixel value of 1 × 5 pixels is input to the edge direction determination unit 4 every clock and an edge direction pos for one pixel is output. The edge direction is a direction from a small pixel value to a large pixel. If the target pixel is the edge of the foreground or background, the direction from the background to the foreground is determined as the edge direction.

  The edge direction determination unit 4 calculates the difference between the pixel of interest (pixel with identification number 25) and the eight neighboring pixels (pixels with identification numbers 17, 18, 19, 24, 26, 31, 32, and 33) adjacent to the pixel of interest. SP and difference SF between the target pixel (pixel of identification number 25) and peripheral pixels (pixels of identification numbers 11, 23, 27, and 39) that are two pixels away from the target pixel in the top, bottom, left, and right of the target pixel are calculated.

The calculation formula is as follows. C is the pixel value of each pixel indicated by the identification number in [].
SP [u] = C [25] -C [18]
SP [l] = C [25] −C [24]
SP [r] = C [25] -C [26]
SP [d] = C [25] -C [32]
SF [u] = C [25] −C [11]
SF [l] = C [25] −C [23]
SF [r] = C [25] −C [27]
SF [d] = C [25] −C [39]

  The edge direction determination unit 4 determines the edge direction x in the main scanning direction and the edge direction y in the sub scanning direction of the pixel of interest. x and y can be 0 or ± 1. 0 indicates that there is no edge, and 1 indicates that there is an edge. Positive and negative signs correspond to directions.

For example, if SP [l] <SP [r], the pixel values of the peripheral pixels on the right side of the left and right of the target pixel are small as shown in FIG. Therefore, the edge direction determination unit 4 determines that the edge direction in the main scanning direction is left, and outputs x = -1. On the other hand, if SP [l]> SP [r], the left pixel value of the left and right of the target pixel is small as shown in FIG. Therefore, the edge direction determination unit 4 determines that the edge direction in the main scanning direction is right, and outputs x = + 1.
Similarly, the edge direction in the sub-scanning direction is determined by comparing SP [u] and SP [d]. If SP [u]> SP [d], the edge direction determination unit 4 determines that the edge direction in the sub-scanning direction is down, and outputs y = -1. If SP [u] <SP [d], the edge direction determination unit 4 determines that the edge direction in the sub-scanning direction is upward, and outputs y = + 1.

  When SP [l] = SP [r], there is no difference in pixel value between the left and right of the target pixel. Similarly, when SP [u] = SP [d], there is no difference in pixel values above and below the target pixel. In this case, the edge direction determination unit 4 determines that there is no edge on the left and right or top and bottom of the target pixel, and outputs x = 0 and y = 0, respectively.

The edge direction determination unit 4 determines the edge direction even when the target pixel is not an edge but is an inner edge one pixel inside the edge.
When the target pixel is the inner edge, the edge direction is x = y = 0. Therefore, when the edge direction is x = y = 0, the edge direction determination unit 4 determines SF [l], SF [r], SF [u], SF obtained from peripheral pixels that are two pixels away from the target pixel. [d] corrects the edge directions x and y.

  The edge direction determination unit 4 corrects x = 0 to x = −1 if SF [l] <SF [r], and corrects x = + 1 if SF [l]> SF [r]. To do. Similarly, if SF [u]> SF [d], the edge direction determination unit 4 corrects y = 0 to y = −1, and if SF [u] <SF [d], y = Correct to +1. If SF [l] = SF [r], no correction is made, and the edge direction determination unit 4 outputs x = 0 as it is. Even in the case of SF [u] = SF [d], no correction is performed, and the edge direction determination unit 4 outputs y = 0 as it is.

  For example, as shown in FIG. 8, when the peripheral pixel on the right side that is two pixels away from the target pixel is an edge and the target pixel is an inner edge, SP [l] = SP [r]. Determined. However, since SF [l] <SF [r], x = 0 is corrected to x = -1.

  The edge direction determination unit 4 determines and outputs a comprehensive edge direction pos by a combination of the edge direction x in the main scanning direction and the edge direction y in the sub scanning direction. pos can take a value of 0-4. 0 indicates the center (no edge), 1 indicates the left direction, 2 indicates the right direction, 3 indicates the upward direction, and 4 indicates the downward direction. The edge direction pos for the combination of x and y is as follows.

When x = 1 and y = −1, pos = 4
When x = 1 and y = 0, pos = 2
When x = 1 and y = + 1, pos = 2
When x = 0 and y = −1, pos = 4
pos = 0 when x = 0 and y = 0
When x = 0 and y = + 1, pos = 3
When x = -1 and y = -1, pos = 1
When x = -1 and y = 0, pos = 1
When x = -1 and y = + 1, pos = 3

The blend processing unit 5 determines the pixel value of each pixel when the thinning process and the smoothing process are realized simultaneously.
The blend processing unit 5 inputs the image data of the 2nd to 6th lines in the line memory for 5 clocks, and obtains 5 × 5 pixels centered on the target pixel of the identification number 25. Further, the blend processing unit 5 shifts the processing result (change amounts dSMT, dST) of 1 × 3 pixels output from the smoothing processing unit 2 and the thinning processing unit 3 by 3 pixels in the main scanning direction, and has an identification number of 25 The processing result for 3 × 3 pixels centering on the target pixel is input. Further, the blend processing unit 5 inputs the edge direction pos of the target pixel from the edge direction determination unit 4. The blend processing unit 5 determines a pixel value G when both thinning and smoothing are simultaneously realized in the target pixel according to the combination of the presence or absence of thinning and the presence or absence of smoothing in the target pixel. The pixel value G is determined by determining the target pixel, the neighboring pixels adjacent to the target pixel in the vertical and horizontal directions, the pixel values C of the peripheral pixels that are two pixels away from the target pixel in the vertical and horizontal directions of the target pixel, Changes dST and dSMT of pixel values of neighboring pixels adjacent to the left and right, and the edge direction pos of the target pixel are used.

As shown in FIG. 9, the blend processing unit 5 includes a classification unit 50 and pixel value determination units 51 to 54.
The classification unit 50 classifies the thinning and smoothing states in the target pixel into cases 1 to 4 according to the combination of the presence or absence of thinning and the presence or absence of smoothing in the target pixel. The presence or absence of thinning and the presence or absence of smoothing can be determined based on whether or not the values of the change amounts dST and dSMT are zero. If the change amount dST [25] input from the thinning processing unit 3 for the target pixel is dST [25] ≠ 0, the classification unit 50 performs thinning on the target pixel as a result of the thinning process. [25] = 0, it is determined that thinning has not been performed. Similarly, the classification unit 50 performs smoothing if the change amount dSMT [25] input from the smoothing processing unit 2 for the target pixel is dSMT [25] ≠ 0, and performs smoothing if dSMT [25] = 0. Judge that it was not done.

FIG. 10 is a flowchart showing the flow of processing by the classification unit 50.
As shown in FIG. 10, the classification unit 50 performs classification into case 1 when the change amount dST [25] ≠ 0 and dSMT [25] ≠ 0 of the target pixel (step S11; Y) (step S12). When the amount of change dST [25] ≠ 0 and dSMT [25] = 0 (step S11; N, S13; Y), the classification unit 50 classifies the pixel of interest into case 2 (step S14). When the change amount dST [25] = 0 and dSMT [25] ≠ 0 of the target pixel (step S13; N, S15; Y), the classification unit 50 classifies the case 3 (step S16). When the change amount dST [25] = 0 and dSMT [25] = 0 of the target pixel (step S15; N, S17; Y), the classification unit 50 classifies the case 4 (step S18).

  The classification unit 50 is classified into the pixel value determination unit 51 when classified into the case 1, the pixel value determination unit 52 when classified into the case 2, and the pixel value determination unit 53 when classified into the case 3. In the case of the case 4 classification, the change amounts dST, dSMT, the pixel value C, and the edge direction pos are output to the pixel value determination unit 54.

  The pixel value determination units 51 to 54 determine the pixel value G [25] of the pixel of interest when both the thinning process and the smoothing process are realized simultaneously according to the cases 1 to 4.

<Case 1>
FIG. 11 is a flowchart of processing executed by the pixel value determination unit 51 for Case 1.
In case 1, since the pixel of interest is thinned and smoothed, it may be a foreground edge as shown in FIG. 12 (a) or a background edge as shown in FIG. 12 (b). is assumed. In the case of the foreground edge, as shown in FIG. 12A, the amounts of change dST and dSMT of the target pixel M are both negative values, and the pixel value is reduced by the thinning process and the smoothing process, respectively. In the case of the background edge, as shown in FIG. 12B, the changes dST and dSMT of the target pixel M are both positive values, and the pixel value is increased by the thinning process and the smoothing process, respectively.

  In any case, the target pixel M is thinned and smoothed, and the peripheral pixel N adjacent to the edge direction pos of the target pixel M or the reverse edge direction apos which is the reverse direction of the edge direction is not thinned or smoothed. Therefore, as shown in FIG. 11, the pixel value determination unit 51 uses a thinning process and a smoothing process to obtain a value obtained by adding the change amounts dST [25] and dSMT [25] to the original pixel value C [25] of the target pixel. The pixel value G [25] of the pixel of interest when processing is realized simultaneously is determined and output (step S21).

  If the determined pixel value G [25] of the target pixel is not between the foreground pixel value and the background pixel value, the density of the target pixel protrudes at the boundary between the foreground and the background, which is unnatural. Therefore, when the pixel value G [25] of the target pixel is larger than the pixel value C [pos] of the surrounding pixels adjacent to the foreground, that is, the edge direction of the target pixel (step S22; Y), the pixel value determination unit 51 is determined. The pixel value G [25] of the target pixel is corrected to the foreground pixel value C [pos] (step S23). Further, when the pixel value G [25] of the target pixel is smaller than the background, that is, the pixel value C [apos] of the peripheral pixel adjacent in the reverse edge direction of the target pixel (step S22; Y), the pixel value determination unit 51 determines The pixel value G [25] of the target pixel thus corrected is corrected to the background pixel value C [apos] (step S23). For apos, a value in the direction opposite to the edge direction pos is set. That is, if pos = 1, apos = 2, and if pos = 3, apos = 4. If there is no edge direction, pos = apos = 0.

<Case 2>
FIG. 13 is a flowchart of processing executed by the pixel value determination unit 52 for Case 2.
In case 2, since the pixel of interest is only thinned, it is assumed that it is a foreground edge as shown in FIG. 14A or a background edge as shown in FIG. In the case of the foreground edge, as shown in FIG. 14A, the change amount dST of the pixel of interest M is a negative value, and the pixel value is reduced by the thinning process. The change amount dSMT of the peripheral pixel N adjacent to the target pixel M in the reverse edge direction apos is a positive value, and the pixel value is increased by the smoothing process. In the case of a background edge, as shown in FIG. 14B, the change amount dST of the target pixel M is a positive value, and the pixel value is increased by the thinning process. The change amount dSMT of the peripheral pixel N adjacent to the target pixel M in the edge direction pos is a negative value, and the pixel value is reduced by the smoothing process.

  In this way, when the smoothed peripheral pixel is adjacent to the edge direction pos or the reverse edge direction apos of the thinned pixel of interest, the pixel value determining unit 52, along with the pixel of interest, includes the fine line of the peripheral pixel. The pixel value when the smoothing process and the smoothing process are realized simultaneously is determined.

  Since the target pixel is only thinned, the pixel value determination unit 52 changes the pixel value change amount dST of the target pixel by the thinning process to the original pixel value C [25] of the target pixel as shown in FIG. The value to which [25] is added is temporarily determined as the pixel value G [25] of the target pixel when the thinning process and the smoothing process are realized simultaneously (step S31). Next, the pixel value determination unit 52 determines whether the smoothed peripheral pixel is adjacent to the edge direction pos or the reverse edge direction apos.

  The pixel value determination unit 52 determines that the amounts of change dST [pos] and dSMT [pos] input for peripheral pixels adjacent to the target pixel in the edge direction pos and edge direction pos are pos ≠ 0, dST [pos] = 0, If dSMT [pos] ≠ 0 is satisfied (step S32; Y), it is determined that the smoothed peripheral pixel is adjacent to the edge direction pos of the target pixel (step S33).

The pixel value determination unit 52 determines that the amounts of change dST [apos] and dSMT [apos] input to the neighboring pixels adjacent to the target pixel in the reverse edge direction apos and the reverse edge direction apos are apos ≠ 0, dST [apos] = When 0 and dSMT [apos] ≠ 0 are satisfied (step S32; N, S34; Y), it is determined that the smoothed peripheral pixel is adjacent to the background, that is, the reverse edge direction apos of the target pixel (step S35).
When none of the conditions is satisfied (step S34; N), the pixel value determination unit 52 determines that the smoothed peripheral pixel is not adjacent to the edge direction pos or the reverse edge direction apos, and is determined in step S31. The pixel value G [25] of the target pixel is output.

  When the direction in which the smoothed neighboring pixels are adjacent to each other is specified, the pixel value determination unit 52 changes the pixel value C [25] of the target pixel to the original pixel value C [25] of the target pixel by the thinning process. Then, a value obtained by adding the pixel value change amount dSMT [P] of the peripheral pixel due to the smoothing process is determined as the pixel value G [25] of the target pixel (step S36). P is input as pos or apos. For dSMT [P], dSMT [pos] is used if the smoothed neighboring pixel is adjacent to the edge direction pos, and dSMT [apos] is used if it is adjacent to the reverse edge direction apos.

  Thereafter, the pixel value G [25] of the target pixel is corrected by the foreground and background pixel values in the same manner as in Case 1. Therefore, the same steps as those in FIG. 11 are denoted by the same step numbers in FIG.

<Case 3>
FIG. 15 is a flowchart of processing executed by the pixel value determination unit 53 for case 3.
Since the pixel of interest is only smoothed in Case 3, the pixel of interest may be a background edge as shown in FIG. 16 (a) or a foreground edge as shown in FIG. 16 (b). is assumed. In the case of the background edge, as shown in FIG. 16A, the amount of change dSMT of the target pixel M is a positive value, and the pixel value is increased by the smoothing process. The change amount dST of the peripheral pixel N adjacent to the target pixel M in the edge direction pos is a negative value, and the pixel value is reduced by the thinning process. In the case of the foreground edge, as shown in FIG. 16B, the amount of change dSMT of the target pixel M is a negative value, and the pixel value is reduced by the smoothing process. The change amount dST of the peripheral pixel N adjacent to the target pixel M in the reverse edge direction apos is a positive value, and the pixel value is increased by the thinning process.

  As described above, when the thinned peripheral pixel is adjacent to the edge direction pos or the reverse edge direction apos of the smoothed target pixel, the pixel value determination unit 53 performs thinning processing and smoothing of the peripheral pixel together with the target pixel. The pixel value G [P] when the processing is realized simultaneously is determined.

  Since the target pixel is only smoothed, the pixel value determining unit 53 adds the target pixel change amount dSMT [25] by the smoothing process to the original pixel value C [25] of the target pixel as shown in FIG. The value is temporarily determined as the pixel value G [25] of the pixel of interest when the thinning process and the smoothing process are realized simultaneously (step S41). Next, the pixel value determination unit 53 determines in which direction of the edge direction pos and the reverse edge direction apos the thinned peripheral pixels are adjacent.

  The pixel value determination unit 53 determines that the amounts of change dST [pos] and dSMT [pos] input for the neighboring pixels adjacent to the target pixel in the edge direction pos and edge direction pos are pos ≠ 0, dST [pos] ≠ 0, When dSMT [pos] = 0 is satisfied (step S42; Y), it is determined that the thinned peripheral pixel is adjacent to the edge direction pos of the target pixel (step S43).

The pixel value determination unit 53 determines that the changes dST [apos] and dSMT [apos] input for the neighboring pixels adjacent to the target pixel in the reverse edge direction apos and the reverse edge direction apos are apos ≠ 0 and dST [apos] ≠. When 0 and dSMT [apos] = 0 are satisfied (step S42; N, S44; Y), it is determined that the thinned peripheral pixel is adjacent to the reverse edge direction apos of the target pixel (step S45).
If neither condition is satisfied (step S44; N), the pixel value determination unit 53 determines that the thinned peripheral pixel is not adjacent to the edge direction pos or the reverse edge direction apos, and is determined in step S41. The pixel value G [25] of the target pixel is output.

When the direction in which the thinned peripheral pixels are adjacent to each other is specified, the pixel value determining unit 53 is larger between the pixel value C [25] of the target pixel and the pixel value C [P] of the thinned peripheral pixel. One is input to L and the smaller one is input to S (step S46), and temp is calculated by the following equation (step S47).
temp = L + S + dST [P] + dSMT [25]
P is input as pos or apos. C [P] and dST [P] are C [pos] and dST [pos] if the thinned peripheral pixels are adjacent in the edge direction pos, and C [P] and dST [P] if they are adjacent in the reverse edge direction. apos], dST [apos].

  “temp” is a sum total of the pixel value of the smoothed target pixel and the pixel values of the thinned peripheral pixels. The pixel value determination unit 53 distributes this temp to the target pixel and the peripheral pixels.

  When the sum of the change amount dSMT [25] of the pixel value of the target pixel due to the smoothing process and the change amount dST [P] of the pixel value of the peripheral pixel due to the thinning process is a positive value, the smoothing of the target pixel and the peripheral pixel The total temp of the pixel values of the two pixels becomes larger than the sum L + S of the original pixel values. In this case, the pixel value determination unit 53 assigns the pixel value L to the pixel with the larger original pixel value, and assigns the remaining pixel value temp-L to the other pixel value. If the total temp is larger than twice L, the pixel value of both the target pixel and the surrounding pixels is equal to or greater than L, and is not adapted to the foreground or the background. In this case, the pixel value determining unit 53 assigns the maximum allowable L value to any of them.

As shown in FIG. 15, when temp> L + S is satisfied (step S48; Y) and temp <L × 2 is satisfied (step S49; Y), the pixel value determining unit 53 determines the original pixel value C [25 of the target pixel. ] Is larger than the original pixel value C [P] of the peripheral pixel, the pixel value G [25] = L of the target pixel and the pixel value G [P] = temp−L of the peripheral pixel are determined. If the original pixel value C [25] of the target pixel is smaller than the original pixel value C [P] of the surrounding pixels, the pixel value determining unit 53 sets the pixel value G [25] = temp−L of the target pixel, The pixel value G [P] = L is determined (step S50).
When temp> L + S is satisfied (step S48; Y), but temp <L × 2 is not satisfied (step S49; N), the pixel value G [25] = L of the target pixel, and the pixel value G [P] of the surrounding pixels = L is determined (step S51).

  On the other hand, when the sum of the change amount dSMT [25] of the pixel value of the target pixel due to the smoothing process and the change amount dST [P] of the pixel value of the surrounding pixels due to the thinning process is a negative value, By thinning the peripheral pixels, the sum temp of the pixel values of these two pixels becomes smaller than the sum L + S of the original pixel values. In this case, the pixel value determination unit 53 assigns the pixel value S to the pixel with the smaller original pixel value, and assigns the remaining pixel value temp-S to the other pixel. Note that if the total temp is smaller than twice S, the pixel values of the target pixel and the surrounding pixels are equal to or less than S, and neither the foreground nor the background is suitable. In this case, the pixel value determination unit 53 assigns the maximum allowable S value to any of them.

As shown in FIG. 15, when temp> L + S is not satisfied (step S48; N), and temp> S × 2 is satisfied (step S52; Y), the original pixel value C [25] of the target pixel is the surrounding pixel. If the pixel value is larger than the original pixel value C [P], the pixel value determination unit 53 determines the pixel value G [25] = temp−S of the target pixel and the pixel value G [P] = S of the peripheral pixels. If the original pixel value C [25] of the pixel of interest is smaller than the original pixel value C [P] of the surrounding pixels, the pixel value determining unit 53 sets the pixel value G [25] = S of the pixel of interest, and the pixel values of the surrounding pixels. G [P] = temp-S is determined (step S53).
If temp> L + S is not satisfied (step S48; N), and temp> S × 2 is not satisfied (step S52; N), the pixel value determination unit 53 determines the pixel value G [25] = S of the target pixel, The pixel value G [P] = S is determined (step S54).

<Case 4>
FIG. 17 is a flowchart of processing executed by the pixel value determination unit 54 for Case 4.
In case 4, since the pixel of interest is neither thinned nor smoothed, the pixel of interest is the inner edge of the foreground as shown in FIG. 18 (a) or the inner edge of the background as shown in FIG. 18 (b). It is assumed that In the case of the inner edge of the foreground, as shown in FIG. 18A, the target pixel M is neither thinned nor smoothed, but the amounts of change dST and dSMT of the peripheral pixel N adjacent to the target pixel M in the reverse edge direction apos are Since each is a negative value, the pixel value is reduced by the thinning process and the smoothing process. In the case of the inner edge of the background, as shown in FIG. 18B, the target pixel M is neither thinned nor smoothed, but the amounts of change dST and dSMT of the peripheral pixel N adjacent to the target pixel M in the edge direction are positive. Therefore, the pixel value is increased by the thinning process and the smoothing process.

  In this way, when the peripheral pixel that has been thinned and smoothed in the edge direction pos or the reverse edge direction apos is adjacent to the target pixel, the pixel value determination unit 54 performs the thinning process and the smoothing process on the peripheral pixel together with the target pixel. Determines the pixel value G [P] at the same time.

  Since the target pixel is neither thinned nor smoothed, the pixel value determination unit 54 realizes the thinning process and the smoothing process simultaneously on the original pixel value C [25] of the target pixel as shown in FIG. Is temporarily determined as the pixel value G [25] of the current pixel of interest (step S61).

  Next, the pixel value determination unit 54 determines whether or not the target pixel, the peripheral pixel adjacent to the edge direction pos of the target pixel, and the peripheral pixel adjacent to the reverse edge direction apos have substantially the same pixel value. (Step S62). This is because when the pixel of interest is the inner edge, the pixel values of the pixel of interest and the peripheral pixels adjacent to the edge direction pos and the reverse edge direction apos are substantially the same. The pixel value determination unit 54 subtracts the pixel value C [apos] of the peripheral pixel adjacent to the reverse edge direction apos from the pixel value C [pos] of the peripheral pixel adjacent to the edge direction pos and the edge value pos. If all of the values obtained by subtracting the pixel value C [25] of the target pixel from the pixel value C [pos] of the surrounding pixel are less than the threshold Th3, it is determined that the pixel values of the three pixels are substantially the same. If either one exceeds the threshold Th3, the pixel value determination unit 54 determines that the pixel values of the three pixels are not substantially the same.

  If the pixel values of the three pixels are not substantially the same (step S62; N), the pixel value determination unit 54 determines the pixel value determined in step S61 because the target pixel is located not on the inner edge but on the inner edge. G [25] is output. When the target pixel is located inside the inner edge, the peripheral pixels adjacent to the target pixel are also located inside the inner edge or the inner edge, so that thinning or smoothing is not appropriate for the peripheral pixel. Therefore, the pixel value determination unit 54 forcibly modifies the amount of change in the surrounding pixels to dST [pos] = dST [apos] = 0 (step S63), and invalidates thinning and smoothing.

  When the pixel values of the three pixels are substantially the same (step S62; Y), the target pixel is the inner edge. The pixel value determining unit 54 determines whether the thinned and smoothed peripheral pixel is adjacent in the edge direction pos or the reverse edge direction apos of the target pixel (step S64).

  The pixel value determination unit 54 determines that the amounts of change dST [pos] and dSMT [pos] input for the neighboring pixels adjacent to the target pixel in the edge direction pos and edge direction pos are pos ≠ 0, dST [pos] ≠ 0, When dSMT [pos] ≠ 0 is satisfied (step S64; Y), it is determined that the thinned and smoothed peripheral pixel is adjacent to the edge direction pos of the target pixel (step S65).

The pixel value determination unit 52 determines that the amounts of change dST [apos] and dSMT [apos] input for the neighboring pixels adjacent to the target pixel in the reverse edge direction apos and the reverse edge direction apos are apos ≠ 0 and dST [apos] ≠. When 0 and dSMT [apos] ≠ 0 are satisfied (step S64; N, S66; Y), it is determined that the thinned and smoothed peripheral pixel is adjacent to the reverse edge direction apos of the target pixel (step S67).
When none of the conditions is satisfied (step S66; N), the pixel value determination unit 54 determines that the thinned and smoothed neighboring pixels are not adjacent to either the edge direction or the reverse edge direction of the target pixel, The pixel value G [25] of the target pixel determined in step S61 is output.

  When it is determined that the thinned and smoothed peripheral pixel is adjacent to the edge direction pos, the pixel value determination unit 54 adds the thinning change amount dST [pos] to the pixel value C [pos] of the peripheral pixel. The value obtained by adding the change amount dSMT [pos] is determined as the pixel value G [pos] of the peripheral pixels when the thinning process and the smoothing process are realized simultaneously (step S68).

  When the pixel value G [pos] of the surrounding pixels becomes larger than the pixel value of the foreground due to thinning and smoothing, it is unnatural. Therefore, the pixel value determination unit 54 determines that the determined pixel value G [pos] is larger than the pixel value C [2pos] of a peripheral pixel that is two pixels away from the target pixel in the edge direction pos (step S69; Y). The pixel value G [pos] of the peripheral pixel is corrected to the pixel value C [2pos] of the peripheral pixel separated by two pixels. Further, the pixel value determination unit 54 adds a value obtained by adding the remaining pixel value G [pos] −C [2pos] that could not be processed by the peripheral pixels to the pixel value C [25] of the target pixel as the thinning process. The pixel value G [25] of the pixel of interest when the smoothing process is realized simultaneously is determined (step S70).

  Since the determined pixel value G [25] of the target pixel is also kept below the foreground pixel value, the determined pixel value G [25] of the target pixel is a pixel value C [2pos] of a peripheral pixel that is two pixels away in the edge direction. ] (Step S71; Y), the pixel value determination unit 54 further modifies the pixel value G [25] of the pixel of interest to the pixel value C [2pos] of the peripheral pixel that is two pixels away (step S72).

  On the other hand, when it is determined that the thinned and smoothed peripheral pixel is adjacent to the reverse edge direction apos, the pixel value determination unit 54 changes the pixel value C [apos] of the peripheral pixel to the change amount dST [ apos] and a value obtained by adding the change amount dSMT [apos] due to the smoothing process are determined as the pixel value G [apos] of the peripheral pixel when the thinning process and the smoothing process are realized simultaneously (step S73).

  When the pixel value G [apos] of the surrounding pixels becomes smaller than the background pixel value due to thinning and smoothing, it is unnatural. Therefore, the pixel value determination unit 54 determines that the determined pixel value G [apos] is smaller than the pixel value C [2apos] of the peripheral pixel that is 2 pixels away from the target pixel in the reverse edge direction (step S74; Y). The pixel value G [apos] of the peripheral pixel is corrected to the pixel value C [2apos] of the peripheral pixel separated by two pixels. In addition, the pixel value determination unit 54 adds the remaining pixel value G [apos] −C [2apos], which could not be processed by the surrounding pixels, to the pixel value C [25] of the target pixel, thereby adding the pixel value G of the target pixel. [25] is determined (step S75).

  Since the determined pixel value G [25] of the target pixel is also kept below the background pixel value, the determined pixel value G [25] of the target pixel is a pixel value C [[ If it is smaller than 2 apos] (step S76; Y), the pixel value determination unit 54 further modifies the pixel value G [25] of the target pixel to the pixel value C [2apos] of the peripheral pixel that is 2 pixels away (step S77).

FIG. 19A is an image diagram of processing performed by the blend processing unit 5.
When thinning processing and smoothing processing are performed on the image shown in FIG. 19A, the pixel of interest M1 is only thinned and is classified as case 2. At this time, since the peripheral pixels adjacent in the edge direction and the reverse edge direction are neither thinned nor smoothed, the pixel value of the pixel of interest M1 that has been thinned only is output.

  On the other hand, the target pixel M2 is smoothed, and the peripheral pixel N2 adjacent in the edge direction of the target pixel M2 is thinned. Accordingly, attention is paid so that the pixel value obtained by adding the pixel value of the pixel of interest M2 classified and smoothed in case 3 and the pixel value of the thinned peripheral pixel N2 to the peripheral pixel N2 having a large original pixel value. It is distributed to the pixel M2 and the peripheral pixel N2.

  The pixel of interest M3 is both smoothed and thinned, and neighboring pixels adjacent in the edge direction or the reverse edge direction are neither thinned nor smoothed. In this case, the pixel value of the pixel of interest M3 which is classified as case 1, thinned, and smoothed is output.

As a result, as shown in FIG. 19B, the straight line portion is thinned, and the hatched portion is smoothly stepped by smoothing, so that thinning and smoothing are realized in a balanced manner.
Considering that such a favorable processing result of the blend processing unit 5 is a thinning process for a straight line portion, a smoothing process for a hatched portion, a method of switching processing depending on the structure of an image, and a thinning process are simultaneously performed. In the method of setting the pixel value changed during the smoothing process to a small value, it is difficult to realize thinning and smoothing at the same time.

FIG. 20A shows the original image. FIG. 20B shows a target image to be realized when the thinning process and the smoothing process are simultaneously performed on the original image.
When only the smoothing process is performed on the original image, as shown in FIG. 20C, the pixel value of the pixel located at the edge of the shaded portion is increased, resulting in a smooth shaded line. When only the thinning process is performed on the original image, as shown in FIG. 20D, the pixel value of the straight line portion is reduced and the line is visually thinned.

When both the smoothing process and the thinning process are performed at the same time, it is expected that the hatched portion becomes smooth and the straight line portion has a reduced edge pixel value as in the target image shown in FIG. However, if switching is performed so that the hatched portion is simply smoothed and the straight line portion is thinned, an unexpected step occurs at the location indicated by the arrow as shown in FIG.
Further, when the pixel value changed during the smoothing process is set to a small value in consideration of the fact that the thinning process is also performed at the same time, the processing result as shown in FIG. Although an improvement is seen from the processing result shown in FIG. 20 (e), a step is generated at the position indicated by the arrow, and the processing result of the target image and the blend processing unit 5 according to the present embodiment (see FIG. 19 (b)). ) Is inferior in quality.

  In the blending process that realizes the thinning process and the smoothing process at the same time, if the thinned pixels are isolated, neither thinning nor smoothing is realized, resulting in image quality deterioration. Therefore, as shown in FIG. 1, the image processing apparatus 1 includes a thinning adjustment unit 6. The thinning adjustment unit 6 invalidates thinning by the thinning processing unit 3 when the thinned pixels are isolated. .

  The thinning adjustment unit 6 receives the change amount dST of the target pixel and its peripheral pixels in the blend processing unit 5 from the thinning processing unit 3, and is thinned at the target pixel and its peripheral pixels by the input change amount dST. It is determined whether or not the pixel is isolated. If it is determined to be isolated, the thinning adjustment unit 6 resets the change dST of the thinned pixel of the target pixel and the peripheral pixels to a value of 0, and invalidates the thinning. The thinning adjustment unit 6 invalidates the thinning according to the control signal SW from the control unit 7.

  The control unit 7 outputs a control signal SW to the thinning adjustment unit 6 and switches whether or not to invalidate the thinning by the thinning adjustment unit 6. The control unit 7 outputs SW = 1 when setting to invalidate thinning, and outputs SW = 0 when setting not to invalidate.

FIG. 21 is a flowchart of processing executed by the thinning adjustment unit 6.
As shown in FIG. 21, the thinning adjustment unit 6 holds the change amount dST of 1 × 3 pixels output from the three thinning processing units 3 for three pixels in the main scanning direction, and the pixel of interest with the identification number 25 And a change amount dST of 3 × 3 pixels composed of the surrounding pixels. The thinning adjustment unit 6 counts the number of pixels thinned by the thinning process among the 3 × 3 pixels, that is, the number of pixels with dST ≠ 0, and outputs the counted number as the degree of isolation CN (step S81). .

  If the isolation degree CN, that is, the number of thinned pixels is 1 or 2, it is considered that the thinned pixels are isolated in 3 × 3 pixels. Hereinafter, an example in which the thinning adjustment unit 6 determines that the thinned pixel is isolated if the isolation level CN is greater than 0 and equal to or less than 2 is used. However, the isolation level CN is used as a criterion for isolating the thinned pixel. A condition that is greater than 0 and equal to or less than 1 may be used. Even if the isolation criterion CN is a criterion of 1 or less, the same effect as in the case of a criterion of 2 or less can be obtained.

  When CN ≦ 2 and SW = 1 (step S82; Y), the thinning adjustment unit 6 is a thin line of the thinned pixel among the target pixel and the peripheral pixel adjacent to the target pixel in the edge direction or the reverse edge direction. The pixel value change amounts dST [25], dST [pos], and dST [apos] due to the conversion process are forcibly reset to 0 (step SS83). When the thinned pixel is a neighboring pixel adjacent to the edge direction pos or the reverse edge direction apos, the thinning adjustment unit 6 changes the amount of change dST [pos of the neighboring pixel on the condition that the target pixel is thinned. ], DST [apos] is preferably reset. After the reset, the thinning adjustment unit 6 outputs a pixel value change amount dST due to the thinning process of 3 × 3 pixels (the target pixel and the surrounding pixels) to the blend processing unit 5.

  When CN ≦ 2 and SW = 1 are not satisfied (step S82; N), either the thinned pixel is not isolated or is set to be isolated but not perform the thinning reset. In any case, the thinning adjustment unit 6 does not reset the change amount dST and outputs the input change amount dST by the thinning process of 3 × 3 pixels to the blend processing unit 5 as it is.

  The classification unit 50 of the blend processing unit 5 performs classification into the above-described cases 1 to 4 using the change amount dST output from the thinning adjustment unit 6. When thinning of the pixel of interest is invalidated, it is classified into case 3 or case 4. In addition, the pixel value determination units 51 to 54 also use the change amount dST output from the thinning adjustment unit 6 to surround G [25] of the pixel of interest, the edge direction pos of the pixel of interest, or the adjacent edge direction apos The pixel values G [pos] and G [apos] of the pixel are determined.

The reproducibility of the fine line is improved by the invalidation of the fine line.
For example, as shown in FIG. 22B, when only the smoothing process is performed on the original image having the oblique line structure in which thin lines having a width of 3 pixels are arranged in a staircase pattern as shown in FIG. Is obtained. The pixel value of each pixel in the step portion is increased or decreased by the smoothing process, and a smooth diagonal line is formed.

  On the other hand, when only the thinning process is performed on the original image, an image as shown in FIG. 22C is obtained as a processing result. As shown in FIG. 22C, when the thinned pixel (the pixel circled in FIG. 22C) is only one pixel and is isolated, the blend processing unit 5 performs the thinning process and the smoothing process. Is simultaneously realized, the pixel value is excessively reduced in the isolated thinned pixels surrounded by a circle in FIG. 22D, and smoothing cannot be realized.

  Therefore, when the thinning of the isolated thinned pixels is invalidated, the image shown in FIG. 22E is obtained as the processing result of the thinning process instead of the image shown in FIG. In FIG. 22E, pixels surrounded by a circle are pixels in which thinning is invalidated. Thereafter, when the thinning process and the smoothing process are realized at the same time by the blend processing unit 5, a fine line image in which the thinning and the smoothing are realized is obtained as shown in FIG.

  As described above, according to the present embodiment, the thinning processing unit 3 that performs thinning processing on image data, the smoothing processing unit 2 that performs smoothing processing on image data, and the edge direction pos for each pixel of the image data. Edge direction determination unit 4 for determining the pixel value of the target pixel, the surrounding pixels of the target pixel, the pixel value change amount dSMT by the smoothing process, the pixel value change amount dST by the thinning process, and the target pixel Blending process for determining the pixel value G [25] of the target pixel when the smoothing process and the thinning process are simultaneously realized according to the combination of the presence or absence of thinning and the presence or absence of smoothing using the edge direction pos Whether or not the thinned pixel is isolated by the amount of change dST of the pixel value due to the thinning processing of the pixel 5 and the pixel of interest and its surrounding pixels If the pixel value is determined to be isolated, the pixel value change amount dST [25 due to the thinning process of the thinned pixel among the target pixel or the peripheral pixels adjacent to the edge direction pos or the reverse edge direction apos of the target pixel is determined. , DST [pos], dST [apos] are reset to 0, and the thinning adjustment unit 6 is provided. The blend processing unit 5 uses the pixel value change amounts dST [25], dST [pos], and dST [apos] due to the reset thinning process to determine the pixel value G [25] of the target pixel.

  Thereby, when realizing the thinning process and the smoothing process at the same time, it is possible to prevent the deterioration of the image quality caused by the isolation of the thinned pixels.

  When the thinned pixel is a neighboring pixel adjacent to the edge direction pos or the reverse edge direction apos of the pixel of interest, the thinning adjustment unit 6 performs a fine line adjustment of the peripheral pixel on the condition that the pixel of interest is thinned. It is preferable to reset the pixel value change amounts dST [pos] and dST [apos] to 0 values.

  Since the control part 7 which switches whether the reset by the thinning adjustment part 6 is performed is provided, it can be switched suitably whether the reset of the variation | change_quantity dST is performed.

The above-described embodiment is a preferred example of the present invention, and the present invention is not limited to this.
For example, the thinning adjustment unit 6 may reset the amount of change dST after the blend processing unit 5 classifies combinations of the presence / absence of thinning and the presence / absence of smoothing in the target pixel into cases 1 to 4. In this case, each of the pixel value determination units 51 to 54 includes the thinning adjustment unit 6, and the processing by the thinning adjustment unit 6 illustrated in FIG. 21 is performed before the processing illustrated in FIGS. 11, 13, 15, and 17. To do. Thereby, in cases 1 and 2, the change amount dST [25] of the target pixel is reset to dST [25] = 0. In cases 3 and 4, the change amount dST [pos] of the neighboring pixels adjacent to the edge direction pos or the change amount dST [apos] of the neighboring pixels adjacent to the reverse edge direction apos is reset to zero.

  In the embodiment described above, since the value of the change amount dST is reset to 0 before the classification of cases 1 to 4, the reset is related to the classification of cases 1 to 4. On the other hand, by performing a reset after classification of cases 1 to 4, the classification of cases 1 to 4 is performed using the original value of change dST, and reset only when the blend processing unit 5 calculates the pixel value. The value of the changed amount dST can be used.

  Further, the processing method of the thinning process and the smoothing process is not limited to the processing method described above. Any processing method may be used as long as the change amount of the pixel value by the thinning process and the change amount of the pixel value by the smoothing process are output.

  Further, although the smoothing processing unit 2 and the thinning processing unit 3 respectively calculated the change amounts dST and dSMT used by the blend processing unit 5, the pixel values after processing are calculated from the smoothing processing unit 2 and the thinning processing unit 3, respectively. And the blend processing unit 5 may calculate the change amounts dST and dSMT by subtracting the original pixel value from the processed pixel value.

  Moreover, the processing content of each part, such as the blend process part 5 and the thinning adjustment part 6, may be programmed, and software processing may be performed in cooperation with this program and the CPU. At this time, as a computer-readable medium for the program, a non-volatile memory such as a ROM and a flash memory, and a portable recording medium such as a CD-ROM can be applied. A carrier wave is also applied to the present invention as a medium for providing program data via a communication line.

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Smoothing process part 3 Thinning process part 4 Edge direction determination part 5 Blend process part 6 Thinning adjustment part 7 Control part

Claims (5)

A thinning processing unit for thinning the image data;
A smoothing processing unit for performing a smoothing process on the image data;
An edge direction determination unit that determines an edge direction for each pixel of the image data;
Using the pixel value of the target pixel, each pixel value of the peripheral pixel of the target pixel, the amount of change of the pixel value by the smoothing process, the amount of change of the pixel value by the thinning process, and the edge direction determined for the target pixel, The blending processing unit that determines the pixel value of the target pixel when the smoothing processing and the thinning processing are realized at the same time according to the combination of the presence or absence of thinning and the presence or absence of smoothing;
It is determined whether or not the thinned pixel is isolated based on the amount of change in the pixel value of the target pixel and its surrounding pixels by the thinning process. If it is determined that the pixel is isolated, the target pixel or the edge direction of the target pixel Or a thinning adjustment unit that resets a change amount of a pixel value due to the thinning process of a thinned pixel among neighboring pixels adjacent in the reverse edge direction to a value of 0,
The blend processing unit uses an amount of change of a pixel value due to the thinning process that is reset to determine a pixel value of the target pixel.   When the thinned pixel is a peripheral pixel adjacent in the edge direction or the reverse edge direction of the target pixel, the thinning adjustment unit thins the peripheral pixel on the condition that the target pixel is thinned. The image processing apparatus according to claim 1, wherein the amount of change in the pixel value due to is reset to a value of zero.   The thinning adjustment unit is configured so that the blending processing unit determines whether there is thinning or smoothing in the pixel of interest according to the amount of change in the pixel value due to the thinning processing of the pixel of interest and the amount of change in the pixel value due to the smoothing processing. The image processing apparatus according to claim 1, wherein the reset is performed after classifying the combinations.   The image processing apparatus according to claim 1, further comprising: a control unit that switches whether to execute the reset by the thinning adjustment unit. A process of thinning the image data;
Applying a smoothing process to the image data;
Determining an edge direction for each pixel of the image data;
Using the pixel value of the target pixel, each pixel value of the peripheral pixel of the target pixel, the amount of change of the pixel value by the smoothing process, the amount of change of the pixel value by the thinning process, and the edge direction determined for the target pixel, The step of determining the pixel value of the pixel of interest when the smoothing process and the thinning process are realized simultaneously according to the combination of the presence or absence of thinning and the presence or absence of smoothing;
It is determined whether or not the thinned pixel is isolated based on the amount of change in the pixel value of the target pixel and its surrounding pixels by the thinning process. If it is determined that the pixel is isolated, the target pixel or the edge direction of the target pixel Or resetting the amount of change in the pixel value due to the thinning process of the thinned pixel among neighboring pixels adjacent in the reverse edge direction to a value of 0,
The step of determining the pixel value of the target pixel is an image processing method for determining the pixel value of the target pixel using the amount of change in the pixel value due to the thinning process that has been reset.

Images